package or namespace load failed for ‘keras’

[max-poltora]

Good afternoon,
I am trying to deploy my shinyApp, which uses keras and imports python modules via reticulate::import function on shynyapps.io. The process seems to run smoothly till the very last moment:

rsconnect::deployApp()
Preparing to deploy application...Update application currently deployed at
https://name.shinyapps.io/app_name/? [Y/n] y
DONE
Uploading bundle for application: 3428026...DONE
Deploying bundle: 4025733 for application: 3428026 ...
Waiting for task: 845012118
  building: Parsing manifest
  building: Building image: 4584803
  building: Installing system dependencies
  building: Fetching packages
  building: Installing packages
  building: Installing files
  building: Pushing image: 4584803
  deploying: Starting instances
  terminating: Stopping old instances
Application successfully deployed to https://name.shinyapps.io/app_name/

But when loaded it gives the following error message:

An error has occurred
The application failed to start: exited unexpectedly with code 1

Error in value[3L] :
package or namespace load failed for ‘keras’:
.onLoad failed in loadNamespace() for 'tensorflow', details:
call: py_module_import(module, convert = convert)
error: ImportError: No module named tools

Detailed traceback:
File "/opt/R/3.6.1/lib/R/library/reticulate/python/rpytools/loader.py", line 24, in _import_hook
level=level
File "/usr/local/lib/python2.7/dist-packages/tensorflow/init.py", line 101, in
from tensorflow_core import *
File "/opt/R/3.6.1/lib/R/library/reticulate/python/rpytools/loader.py", line 24, in _import_hook
level=level
File "/usr/local/lib/python2.7/dist-packages/tensorflow_core/init.py", line 40, in
from tensorflow.python.tools import module_util as _module_util
File "/opt/R/3.6.1/lib/R/library/reticulate/python/rpytools/loader.py", line 24, in _import_hook
level=level
Calls: local ... tryCatch -> tryCatchList -> tryCatchOne ->
Execution halted

The errors message seems to be strange to me, as I do not import any modules called "tools". Here is a peace of my code, that seems to cause execution failure:

library(shiny)
library(shinythemes)
Sys.setenv(RETICULATE_PYTHON = "C:/Users/user/anaconda3/envs/tf")
library(keras)
use_condaenv("tf")

np <- import("numpy", convert=FALSE)
ndi <- import("scipy.ndimage", convert=FALSE)
segment <- import("skimage.segmentation", convert=FALSE)
feature <- import("skimage.feature", convert=FALSE)

model = load_model_hdf5("model_v02122020.h5")

I have found some similar issues (like [this] one (#268)), but suggested solutions do not have an effect in my situation. Could you please kindly help me with this issue?

[max-poltora]

I have slightly modified my code:

reticulate::virtualenv_create(envname = 'tf', python = 'python36')
reticulate::virtualenv_install('tf', c('numpy', 'tensorflow', 'keras'), ignore_installed = FALSE)
reticulate::use_virtualenv(virtualenv = 'tf', required = TRUE)

library(shiny)
library(shinythemes)
library(keras)

np <- import("numpy", convert=FALSE)
ndi <- import("scipy.ndimage", convert=FALSE)
segment <- import("skimage.segmentation", convert=FALSE)
feature <- import("skimage.feature", convert=FALSE)

model = load_model_hdf5("model_v02122020.h5")

But now, the error message is not informative at all. Thus, deployment process runs normally:

rsconnect::deployApp()
Preparing to deploy application...Update application currently deployed at
https://name.shinyapps.io/app_name/? [Y/n] y
DONE
Uploading bundle for application: 3428026...DONE
Deploying bundle: 4025913 for application: 3428026 ...
Waiting for task: 845033169
  building: Parsing manifest
  building: Building image: 4584992
  building: Installing system dependencies
  building: Fetching packages
  building: Installing packages
  building: Installing files
  building: Pushing image: 4584992
  deploying: Starting instances
  rollforward: Activating new instances
  terminating: Stopping old instances
Application successfully deployed to https://name.shinyapps.io/app_name/

The error message:

An error has occurred
The application failed to start: exited unexpectedly with code 1

Error in value[3L] : error in running command
Calls: local ... tryCatch -> tryCatchList -> tryCatchOne ->
Execution halted

[t-kalinowski]

Hello,

Please try putting this in your global.R file:

if(!file.exists(reticulate::miniconda_path()))
  reticulate::install_miniconda()
if(!reticulate::py_module_available("tensorflow"))
  keras::install_keras()
keras::install_keras()

[github-actions]

Automatically closed because there has not been a response for 30 days. When you're ready to work on this further, please comment here and the issue will automatically reopen.