Artificial artifact limit in log_table [BUG]

[marcosjt7]

Issues Policy acknowledgement

• I have read and agree to submit bug reports in accordance with the issues policy

Where did you encounter this bug?

Local machine

Willingness to contribute

Yes. I can contribute a fix for this bug independently.

MLflow version

• Client: 2.8.1
• Tracking server: 2.8.1

System information

• OS Platform and Distribution: MacOS Sonoma 14.4.1
• Python version: 3.10.13

Describe the problem

Unlike the other mlflow.log_* functions, mlflow.log_table internally appends to an mlflow tag mlflow.loggedArtifacts (code).
Since mlflow tags have a character limit of 5000, this creates a limit on the number of table artifacts that can be logged to a single run: If the the sum length of logged artifact paths exceeds 5000, log_table will raise RestException: INVALID_PARAMETER_VALUE.

For now, one can get around this by writing out the data and using mlflow.log_artifact instead. My primary concern is that this pattern of tracking artifacts with a tag of finite length does not get extended to log_artifact.

Tracking information

REPLACE_ME

Code to reproduce issue

import mlflow

with mlflow.start_run() as mlflow_run:
    for i in range(80):
        artifact_path = 'test_directory/this_is_my_test_sub_directory/test_table_{}.csv'.format(i)
        mlflow.log_table(df, artifact_path)

Stack trace

---------------------------------------------------------------------------
RestException                             Traceback (most recent call last)
Cell In[68], line 10
      8 for i in range(80):
      9     artifact_path = 'test_directory/this_is_my_test_sub_directory/test_table_{}.csv'.format(i)
---> 10     mlflow.log_table(df, artifact_path)

File /venv/lib/python3.10/site-packages/mlflow/tracking/fluent.py:1244, in log_table(data, artifact_file)
   1202 """
   1203 Log a table to MLflow Tracking as a JSON artifact. If the artifact_file already exists
   1204 in the run, the data would be appended to the existing artifact_file.
   (...)
   1241         mlflow.log_table(data=df, artifact_file="qabot_eval_results.json")
   1242 """
   1243 run_id = _get_or_start_run().info.run_id
-> 1244 MlflowClient().log_table(run_id, data, artifact_file)

File /venv/lib/python3.10/site-packages/mlflow/tracking/client.py:1611, in MlflowClient.log_table(self, run_id, data, artifact_file)
   1609 current_tag_value.append(tag_value)
   1610 # Set the tag with the updated list
-> 1611 self.set_tag(run_id, MLFLOW_LOGGED_ARTIFACTS, json.dumps(current_tag_value))

File /venv/lib/python3.10/site-packages/mlflow/tracking/client.py:924, in MlflowClient.set_tag(self, run_id, key, value, synchronous)
    870 def set_tag(
    871     self, run_id: str, key: str, value: Any, synchronous: bool = True
    872 ) -> Optional[RunOperations]:
    873     """
    874     Set a tag on the run with the specified ID. Value is converted to a string.
    875 
   (...)
    922         Tags: {'nlp.framework': 'Spark NLP'}
    923     """
--> 924     return self._tracking_client.set_tag(run_id, key, value, synchronous=synchronous)

File /venv/lib/python3.10/site-packages/mlflow/tracking/_tracking_service/client.py:366, in TrackingServiceClient.set_tag(self, run_id, key, value, synchronous)
    364 tag = RunTag(key, str(value))
    365 if synchronous:
--> 366     self.store.set_tag(run_id, tag)
    367 else:
    368     return self.store.set_tag_async(run_id, tag)

File /venv/lib/python3.10/site-packages/mlflow/store/tracking/rest_store.py:234, in RestStore.set_tag(self, run_id, tag)
    225 """
    226 Set a tag for the specified run
    227 
    228 :param run_id: String ID of the run
    229 :param tag: RunTag instance to log
    230 """
    231 req_body = message_to_json(
    232     SetTag(run_uuid=run_id, run_id=run_id, key=tag.key, value=tag.value)
    233 )
--> 234 self._call_endpoint(SetTag, req_body)

File /venv/lib/python3.10/site-packages/mlflow/store/tracking/rest_store.py:59, in RestStore._call_endpoint(self, api, json_body)
     57 endpoint, method = _METHOD_TO_INFO[api]
     58 response_proto = api.Response()
---> 59 return call_endpoint(self.get_host_creds(), endpoint, method, json_body, response_proto)

File /venv/lib/python3.10/site-packages/mlflow/utils/rest_utils.py:210, in call_endpoint(host_creds, endpoint, method, json_body, response_proto, extra_headers)
    208     call_kwargs["json"] = json_body
    209     response = http_request(**call_kwargs)
--> 210 response = verify_rest_response(response, endpoint)
    211 js_dict = json.loads(response.text)
    212 parse_dict(js_dict=js_dict, message=response_proto)

File /venv/lib/python3.10/site-packages/mlflow/utils/rest_utils.py:142, in verify_rest_response(response, endpoint)
    140 if response.status_code != 200:
    141     if _can_parse_as_json_object(response.text):
--> 142         raise RestException(json.loads(response.text))
    143     else:
    144         base_msg = (
    145             f"API request to endpoint {endpoint} "
    146             f"failed with error code {response.status_code} != 200"
    147         )

RestException: INVALID_PARAMETER_VALUE: Tag value '[{"path": "test_directory/this_is_my_test_sub_directory/test_table_0.csv", "type": "table"}, {"path": "test_directory/this_is_my_test_sub_directory/test_table_1.csv", "type": "table"}, {"path": "test_directory/this_is_my_test_sub_directory/test_table' had length 5012, which exceeded length limit of 5000

Other info / logs

REPLACE_ME

What component(s) does this bug affect?

• area/artifacts: Artifact stores and artifact logging
• area/build: Build and test infrastructure for MLflow
• area/deployments: MLflow Deployments client APIs, server, and third-party Deployments integrations
• area/docs: MLflow documentation pages
• area/examples: Example code
• area/model-registry: Model Registry service, APIs, and the fluent client calls for Model Registry
• area/models: MLmodel format, model serialization/deserialization, flavors
• area/recipes: Recipes, Recipe APIs, Recipe configs, Recipe Templates
• area/projects: MLproject format, project running backends
• area/scoring: MLflow Model server, model deployment tools, Spark UDFs
• area/server-infra: MLflow Tracking server backend
• area/tracking: Tracking Service, tracking client APIs, autologging

What interface(s) does this bug affect?

• area/uiux: Front-end, user experience, plotting, JavaScript, JavaScript dev server
• area/docker: Docker use across MLflow's components, such as MLflow Projects and MLflow Models
• area/sqlalchemy: Use of SQLAlchemy in the Tracking Service or Model Registry
• area/windows: Windows support

What language(s) does this bug affect?

• language/r: R APIs and clients
• language/java: Java APIs and clients
• language/new: Proposals for new client languages

What integration(s) does this bug affect?

• integrations/azure: Azure and Azure ML integrations
• integrations/sagemaker: SageMaker integrations
• integrations/databricks: Databricks integrations

[github-actions]

@mlflow/mlflow-team Please assign a maintainer and start triaging this issue.