Tuning parameter by random tune in Caret

I am a new learner for machine learning and R caret package. I have two sets of data to feed into my SVM model for training. I set the tunelength to 9 for optimization. I checked the saved result for both models but found for one of the models, it printed:

ROC was used to select the optimal model using the largest value. The final values used for the model were sigma = 0.004947487 and C = 47.93542.

but for the other one, it printed: Tuning parameter 'sigma' was held constant at a value of 0.003224873 ROC was used to select the optimal model using the largest value. The final values used for the model were sigma = 0.003224873 and C = 8.

The two models are identical except for the input data. I am totally confused about why they are different. Can someone kindly help me with it? Million thanks!