Cloudml package `Error in gsutil_binary() : failed to find 'gsutil' binary`

keras
cloudml

#1

I’m getting the following error when I try to submit a training job to Cloud ML using the cloudml package (I’m replicating the keras fraud example here):

Error in gsutil_binary() : failed to find 'gsutil' binary

I’ve used gcloud_install() and installed everything, and set up my configurations with the shell as it opens after install. Running gsutil --version gives me gsutil version: 4.28, so it seems to have installed OK.

Anything I’m missing?


#2

Same issue here. Using Windows 10, RStudio 1.0.136 and R version 3.3.2. Gsutil seems to have installed OK just as the OP said.


#3

Can you please tell me about your python installation?

Specifically, do you use Anaconda?


#4

Hi @_andrie, I don’t use Anaconda. My Python info is:

$ python --version
Python 3.6.3

#5

This issue has now been fixed. See the issue tracker at https://github.com/rstudio/cloudml/issues/122.

If you pull from the github master branch now, it should work. (If you still experience problems, please report here or file a new issue at https://github.com/rstudio/cloudml/issues)


#6

Great, solved now, thanks @andrie. I still needed an explicit job_collect() command (the Y didn’t work when I was given the choice in the R console), but otherwise worked great.