Recreation model server connection error

Software Support
1

Hello, I am trying to run the InVEST Recreation model (version 3.14.2) and I keep getting a connection error to the remote server. I have tried the following:

  • Running on university WiFi network

  • Running on mobile hotspot

  • Disabling Windows Defender Firewall completely for both private and public networks

In all cases I get the same error: Pyro4.errors.CommunicationError: cannot connect to ('34.44.144.58', 54321)

The elapsed time increased to 2m 21s with firewall disabled, suggesting my computer is reaching out but the server is not responding. Could you please check if the server is currently operational?

InVEST version: 3.14.2 OS: Windows Date/time of attempts: May 26, 2026, ~15:19–15:36 (Chile time, UTC-4)

Thank you,

2

Hi Isabel,

Thanks for reaching out! The Rec server no longer supports versions of InVEST earlier than 3.15.0, so unfortunately you won’t be able to connect while running 3.14.2. If possible, I’d recommend updating to 3.19.0 (the latest version), which includes some improvements to the way the Rec model handles file transfers between the server and the client.

3

Thank you. I will try with the newest version, but is there no access to data between 2010 and 2017, right?

4

No, the newest version still uses the same Flickr data. In version 3.15.0, Twitter data was added as well.

5

Thank you, Megan.

I successfully updated to InVEST 3.19.0 and the model connected to the server. However, after running for 2 hours and 5 minutes the connection was lost with the following error:

ConnectionClosedError: connection lost: [WinError 10054] 
An existing connection was forcibly closed by the remote host

The log shows TUD completed successfully (100%) but PUD was still processing when the connection dropped. No output rasters were generated in the workspace. I was running with cell_size=12.5 and grid_aoi=True over an urban AOI. Should I increase the cell size to avoid timeout? Is there a recommended maximum cell count for the server?

Additionally, I noticed from the log that the Flickr server supports year queries for 2005-2017, while the Twitter server supports year queries for 2012-2022. Since these two sources cover different time ranges, what year range would you recommend using to best combine both datasets? Should I run the full available range for each source separately, or is it better to use a common overlapping period (2012–2017) to ensure comparability between PUD and TUD results?"