Hi, I get the following error when I enable model serving for spacy model via MLFLOW.

[harish_s]

+ echo 'GUNICORN_CMD_ARGS=--timeout 63 --workers 4 '

GUNICORN_CMD_ARGS=--timeout 63 --workers 4 

+ mlflow models serve --no-conda -m /tmp/tmp1a4ltdrk/spacymodelv1 -h unix:/tmp/3.sock -p1

2022/03/01 08:26:37 INFO mlflow.models.cli: Selected backend for flavor 'python_function'

2022/03/01 08:26:37 INFO mlflow.pyfunc.backend: === Running command 'gunicorn --timeout=60 -b unix:/tmp/3.sock:1 -w 1 ${GUNICORN_CMD_ARGS} -- mlflow.pyfunc.scoring_server.wsgi:app'

[2022-03-01 08:26:38 +0000] [24589] [INFO] Starting gunicorn 20.1.0

[2022-03-01 08:26:38 +0000] [24589] [INFO] Listening at: unix:/tmp/3.sock:1 (24589)

[2022-03-01 08:26:38 +0000] [24589] [INFO] Using worker: sync

[2022-03-01 08:26:38 +0000] [24591] [INFO] Booting worker with pid: 24591

[2022-03-01 08:26:38 +0000] [24592] [INFO] Booting worker with pid: 24592

[2022-03-01 08:26:38 +0000] [24593] [INFO] Booting worker with pid: 24593

[2022-03-01 08:26:38 +0000] [24594] [INFO] Booting worker with pid: 24594

[2022-03-01 08:26:40 +0000] [24593] [ERROR] Exception in worker process

Traceback (most recent call last):

 File "/databricks/conda/envs/model-3/lib/python3.8/site-packages/gunicorn/arbiter.py", line 589, in spawn_worker

  worker.init_process()

 File "/databricks/conda/envs/model-3/lib/python3.8/site-packages/gunicorn/workers/base.py", line 134, in init_process

  self.load_wsgi()

 File "/databricks/conda/envs/model-3/lib/python3.8/site-packages/gunicorn/workers/base.py", line 146, in load_wsgi

  self.wsgi = self.app.wsgi()

 File "/databricks/conda/envs/model-3/lib/python3.8/site-packages/gunicorn/app/base.py", line 67, in wsgi

  self.callable = self.load()

 File "/databricks/conda/envs/model-3/lib/python3.8/site-packages/gunicorn/app/wsgiapp.py", line 58, in load

  return self.load_wsgiapp()

 File "/databricks/conda/envs/model-3/lib/python3.8/site-packages/gunicorn/app/wsgiapp.py", line 48, in load_wsgiapp

  return util.import_app(self.app_uri)

 File "/databricks/conda/envs/model-3/lib/python3.8/site-packages/gunicorn/util.py", line 359, in import_app

  mod = importlib.import_module(module)

 File "/databricks/conda/envs/model-3/lib/python3.8/importlib/__init__.py", line 127, in import_module

  return _bootstrap._gcd_import(name[level:], package, level)

 File "<frozen importlib._bootstrap>", line 1014, in _gcd_import

 File "<frozen importlib._bootstrap>", line 991, in _find_and_load

 File "<frozen importlib._bootstrap>", line 975, in _find_and_load_unlocked

 File "<frozen importlib._bootstrap>", line 671, in _load_unlocked

 File "<frozen importlib._bootstrap_external>", line 783, in exec_module

 File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed

 File "/databricks/conda/envs/model-3/lib/python3.8/site-packages/mlflow/pyfunc/scoring_server/wsgi.py", line 6, in <module>

  app = scoring_server.init(load_model(os.environ[scoring_server._SERVER_MODEL_PATH]))

 File "/databricks/conda/envs/model-3/lib/python3.8/site-packages/mlflow/pyfunc/__init__.py", line 710, in load_model

  model_impl = importlib.import_module(conf[MAIN])._load_pyfunc(data_path)

 File "/databricks/conda/envs/model-3/lib/python3.8/site-packages/mlflow/pyfunc/model.py", line 269, in _load_pyfunc

  python_model = cloudpickle.load(f)

 File "stringsource", line 6, in spacy.pipeline.trainable_pipe.__pyx_unpickle_TrainablePipe

_pickle.PickleError: Incompatible checksums (68673003 vs 0x61fbab5 = (cfg, model, name, scorer, vocab))

[2022-03-01 08:26:40 +0000] [24593] [INFO] Worker exiting (pid: 24593)

[2022-03-01 08:26:41 +0000] [24594] [ERROR] Exception in worker process

[harish_s]

Hi @Kaniz Fatma​ 

Thanks for the reply, the version of the Spacy installed is 3.1.3.

Please find the attached details for the same:

 

I will be very thankful, if I get some help here.

[jose_gonzalez]

Which DBR version are you using? also, have you check if the DBR has this library/version?

[Prabakar · Accepted solution]

Hi @Harish S​ this error could happen if the backend services are not updated. Are you doing this test in a PVC environment or a standard workspace?