Does anyone use Xgboost and sparklyr together?

I am trying to deploy xgboost model through sparklyr::spark_apply, it seems not stable.
Need more help.

Related issue: https://github.com/rstudio/sparklyr/issues/362