Execution fails in Docker mode when there is no connection to the Internet

[konstjar]

• Docker mode enabled
• Docker image of the R runtime was imported in advance
• There is no Internet available on the host
• When analysis starts, the code tries to pull image as new and throws the error.

Expectations:

• If image exists and pull fails, analysis should be started on existing image.

[ablack3]

see https://github.com/OHDSI/ArachneExecutionEngine/blob/81ce125f5b44dafda4c87bd2fbc3acb2b8477a7f/engine/src/main/java/com/odysseusinc/arachne/executionengine/execution/r/DockerService.java#L146

Offline execution is a requirement for Darwin

[ablack3]

Steps to reproduce:

Create a text file called main.R with one line print('it works')

Put this file in a folder and zip that folder

Uploade the zip file to Arachne datanode. Select the default runtime.

Before you submit your analysis turn off internet access on your computer.

then try to run the analysis.

I actually am not even able to select the data source when my computer is disconected from the internet. I think that perhaps this is because Arachne tries to check the database which is on another machine.

In general it would be nice if the UI gave a more feedback about what is going on in the underlying application. When things fail it is not at all clear to user. I can look at the docker logs and see the issue but the UI just freezes up in a lot of cases.

This app seems to require internet access in at least a few places. I don't fully understand when internet access is needed or being used and as a user the app just seems to freeze up when there is a problem.

[konstjar]

Resolved in v2.1.0