Shutting down h2o gracefully

[Aschyle]

Hi,

I'm currently running an R script that instantiates an h2o cluster on a Windows R2 2012 server and trains some model

and after successful training, I shutdown h2o with 

h2o.shutdown(F)

after instanciating it via 

h2o.init(ip = "localhost", nthreads = 6, max_mem_size = "18g")

For the sake of completeness, let me add that the script is part of a larger Luigi pipeline that is launched by executing 

a python script from the command line. The shutdown line is the very last line of my R script but I keep getting the following 

error messages in the logs:

2016-07-26 10:46:35.812 - INFO - Shutting down H2O cluster

[1] TRUE

2016-07-26 10:46:35.856 - SIMPLEERROR - No active connection to an H2O cluster. Did you run `h2o.init()` ?

Error in h2o.getConnection() :

  No active connection to an H2O cluster. Did you run `h2o.init()` ?

Calls: <Anonymous> ... .h2o.doREST -> stopifnot -> is -> h2o.getConnection

2016-07-26 10:46:35.870 - SIMPLEERROR - No active connection to an H2O cluster. Did you run `h2o.init()` ?

Error in h2o.getConnection() :

  No active connection to an H2O cluster. Did you run `h2o.init()` ?

Calls: <Anonymous> ... .h2o.doREST -> stopifnot -> is -> h2o.getConnection

This doesn't have any impact on my downstream tasks but it's a little annoying and has never happened by 

running a similar script via source('..'.) from RStudio.

I'm running H2O 3.8.2.3 on a Windows Server (2012 R2) via R 3.3.1 (64bit). I have planned to upgrade the

h2o version in the near future but I'm sticking to this release in order to compare models built with this one.

Any input on this would be appreciated !

[mvh]

I encountered this issue and was able to resolve it by removing all objects and calling garbage collection before shutting down H2O.

rm(list = ls())

gc()

h2o::h2o.shutdown(FALSE)