xgb.cv creates Warning: Empty dataset at worker: 0

Question

I am using a XGBClassifier and try to do a grid search in order to tune some parameters, and I get this warning : WARNING: ../src/learner.cc:1517: Empty dataset at worker: 0 whenever I launch the command xgb.cv(). It does not work either when I try this on default parameters.

Can anybody help? I am kind of lost there!

Here is an example code that produces the warning when I launch it:

import pandas as pd
import xgboost as xgb
from xgboost import cv
import numpy as np

seed = 10
nfold = 10

X = pd.DataFrame(np.random.randint(0,100,size=(6, 8)), columns=list("ABCDEFGH"))
y = pd.Series(np.array(\[0,0,0,1,1,1\]), index = \[0,1,2,3,4,5\])
X_test = pd.DataFrame(np.random.randint(0,100,size=(2, 8)), columns=list("ABCDEFGH"))
X_test.index= \[6,7\]
y_test = pd.Series(np.array(\[0,1\]), index = \[6,7\])

    dtrain = xgb.DMatrix(X, label=y)
    dtest = xgb.DMatrix(X_test, label = y_test)
    
    # Here, this part is just to confirm we get a better model: this model will probably overfit.
    params = {'objective':'multi:softmax'}
    params['eval_metric'] = "mlogloss"
    params['num_class'] = np.unique(y).size # Count how many levels in the response variable.
    num_boost_round = 999
    
    default_model = xgb.train(
        params,
        dtrain,
        num_boost_round=num_boost_round,
        evals=[(dtest, "Test")],
        early_stopping_rounds=10)
    
    cv_results = xgb.cv(
        params,
        dtrain,
        num_boost_round=num_boost_round,
        seed=seed,
        nfold=nfold,
        metrics={'mlogloss'},
        early_stopping_rounds=10)

Scald