Is it possible for both recall and precision to be zeros?

I am trying to evaluate the model performance but I get zeros in one class for both precision and recall (the data is imbalanced with multiple classes > 20 class)

so , Is it possible for both recall and precision to be zeros on test data?

Sure, why not. That just means there are no true positives for this class. Consider this simplified example:

from sklearn.metrics import confusion_matrix, classification_report

y_true = [0,0,0,1,0,0,0,1,0,1]
y_pred = [0,0,0,0,1,0,0,0,1,0]

confusion_matrix(y_true, y_pred)

>>> array([[5, 2],
>>>        [3, 0]])

print(classification_report(y_true, y_pred))

>>>              precision    recall  f1-score   support

>>>        0       0.62      0.71      0.67         7
>>>        1       0.00      0.00      0.00         3

>>> accuracy                           0.50        10
>>> macro avg      0.31      0.36      0.33        10
>>> weighted avg   0.44      0.50      0.47        10