spaCy throws OSError when deployed to AWS Lambda using zappa

Question

When deploying a Python spaCy app to AWS Lambda, I get the following error in the deploy (see below). Why deploy using zappa? The zip file is 125MB compressed, so a direct upload from the aws-cli fails on space, and a transfer to S3 also fails because the uncompressed is more than 250MB.

My program, itself, is not doing any multithreading nor multiprocessing, and it is only using spaCy 2.0. I built and deployed on an EC2 AWS Linux t2.medium. What are the exact steps that get a round-trip answer from a spaCy AWS Lambda function?

Failure trace below:

[1520570028387] Failed to find library...right filename?
[1520570029826] [Errno 38] Function not implemented: OSError
Traceback (most recent call last):
  File "/var/task/handler.py", line 509, in lambda_handler
  return LambdaHandler.lambda_handler(event, context)
  File "/var/task/handler.py", line 237, in lambda_handler
  handler = cls()
  File "/var/task/handler.py", line 129, in __init__
  self.app_module = importlib.import_module(self.settings.APP_MODULE)
  File "/var/lang/lib/python3.6/importlib/__init__.py", line 126, in import_module
  return _bootstrap._gcd_import(name[level:], package, level)
  File "<frozen importlib._bootstrap>", line 978, in _gcd_import
  File "<frozen importlib._bootstrap>", line 961, in _find_and_load
  File "<frozen importlib._bootstrap>", line 950, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 655, in _load_unlocked
  File "<frozen importlib._bootstrap_external>", line 678, in exec_module
  File "<frozen importlib._bootstrap>", line 205, in _call_with_frames_removed
  File "/tmp/spaciness/front.py", line 1, in <module>
  import spacy
  File "/tmp/spaciness/spacy/__init__.py", line 4, in <module>
  from .cli.info import info as cli_info
  File "/tmp/spaciness/spacy/cli/__init__.py", line 1, in <module>
  from .download import download
  File "/tmp/spaciness/spacy/cli/download.py", line 10, in <module>
  from .link import link
  File "/tmp/spaciness/spacy/cli/link.py", line 7, in <module>
  from ..compat import symlink_to, path2str
  File "/tmp/spaciness/spacy/compat.py", line 11, in <module>
  from thinc.neural.util import copy_array
  File "/tmp/spaciness/thinc/neural/__init__.py", line 1, in <module>
  from ._classes.model import Model
  File "/tmp/spaciness/thinc/neural/_classes/model.py", line 12, in <module>
  from ..train import Trainer
  File "/tmp/spaciness/thinc/neural/train.py", line 7, in <module>
  from tqdm import tqdm
  File "/tmp/spaciness/tqdm/__init__.py", line 1, in <module>
  from ._tqdm import tqdm
  File "/tmp/spaciness/tqdm/_tqdm.py", line 53, in <module>
  mp_lock = mp.Lock()  # multiprocessing lock
  File "/var/lang/lib/python3.6/multiprocessing/context.py", line 67, in Lock
  return Lock(ctx=self.get_context())
  File "/var/lang/lib/python3.6/multiprocessing/synchronize.py", line 163, in __init__
  SemLock.__init__(self, SEMAPHORE, 1, 1, ctx=ctx)
  File "/var/lang/lib/python3.6/multiprocessing/synchronize.py", line 60, in __init__
  unlink_now)
OSError: [Errno 38] Function not implemented

Rene B. · Answer 1 · 2019-06-12T15:04:07.007

I could solve the issue with the following steps:

Increase memory size of the lambda function in the zappa_settings.json:

{ "dev": {
```
    "memory_size": 3008,
}
```
}
I had to use a newer version of tqdm. Per default it was version 4.19 which had these issues as described here: https://github.com/tqdm/tqdm/issues/466

The described issue is fixed in a newer version. Its only to add tqdm to my requirements.txt and execute a pip upgrade of the package:

pip install -U tqdm

When I execute zappa deploy dev I get now the following message:

(tqdm 4.32.1 (/var/task/ve/lib/python3.6/site-packages), Requirement.parse('tqdm==4.19.1'), {'zappa'})

tqdm 4.19.1 was the default version of zappa and tqdm 4.32.1 is the new version containing the fix.

spaCy throws OSError when deployed to AWS Lambda using zappa

1 Answers1