AWS Hosting - State File Issue

[nikola-derm]

Hello @vnijs,

We hosted your app on AWS (EC2 instance) but we encountered an issue with state file. When we download the state file to the local machine and re-upload the same state file (using both local and server upload), Radiant does not go back to that state but rather just shows the default two datasets.

Also when clicking on download state file (server), nothing is displayed to the user. Not sure if there is anything happening in the backend.

I apologize as this is probably some misconfiguration on our end but any feedback/help would be helpful.

Thank you very much!

[vnijs]

Hmmm. I'm not able to reproduce any issues like this.

For download server to work you would need to set up an option for the app users to access the server file-system. That can be risky of course.

Can you share an example statefile perhaps?

[nikola-derm]

Thank you for your quick response!

I might allow of server file-system access as this will be internal app. But good point, it would be risky.

Please find an example attached:
jovyan.state-4.rda.zip

It does seem that download state works, if we upload the state file to a locally running Radiant server, the state updates correctly. So only the hosted server has some issue updating to the required state.

Thank you!

[vnijs]

I think I might know what the issue is with the server. Shiny server since a while does this auto reconnect thing that makes restoring state in Radiant impossible. The fix is to set reconnect false; as shown in the exert from our shiny config file below.

  location /radiant {
    
     ....

    utilization_scheduler 10 .5 10;

    site_dir /home/shiny/radiant/inst/app;
    reconnect false;
    log_dir /home/shiny/shiny-server/radiant;

    directory_index off;
  }

[nikola-derm]

I checked the config and this was set by default (reconnect false) for the root path which is the one being used.

[vnijs]

Ah. Then I'm not sure what the issue could be. Saving and restoring state works fine on local and our server so I'm not sure how to reproduce the issue you are seeing. Do the shiny server logs contain any information that might be useful perhaps?

[nikola-derm]

Nothing that looked as a cause for this issue. Thank you for trying to help. If you do not mind I would leave this issue open as I continue to debug in case I find the issue and I can respond here for others to see.

[vnijs]

Of course @nikola-derm. Not a problem