matt (support) wrote:It is odd that having too short of an interval (< 2 minutes?) causes errors with the reflector connection failing. What errors are shown in the Event Log (both from the script you are running and Indigo Server trying to recreate the reflector connection)? I say it is odd because leaving a Web browser up will cause traffic ever few seconds and it didn't sound like you had problems in that scenario. I don't know what would be different from the browser making an HTTP/S call (via JavaScript) versus your Python script making an HTTP/S call.
When the timing gets down in the range of 5-10 seconds, it takes multiple reconnection attempts, sometimes reaching the 15 minute level. The server logs the error in the usual way, showing the connection test failed and the reconnection scheduled after a specified time. The script shows "Reflector connection is currently down" at whatever interval it's using while the reflector is down. I agree it's odd that the script would do this but the browser would not. I can confidently report it but I can't explain it. Some of the inconsistencies may have to do with the relative timing of the reflector checks and the script executions. Also, the browser is sending a few device status requests whereas the script is sending a prismstatus message. I'll try changing the url in the script and browser to see if that changes anything. (Edit: Whoops, I couldn't make them equivalent since the prismstatus url doesn't repeat in the browser and the control page url won't authenticate in the script.)
matt (support) wrote:This really isn't a problem we've had with anyone else that hasn't been resolved except the discussion on this thread, so it doesn't really move the needle on the frequent-problems list. I think maybe we've had 3 or 4 other cases of reflector connectivity issues in the past, but those were solved with changing router settings, replacing the router, or in one case an Indigo change we made because a satellite ISP would totally block the reflector connection request (it would never work because they blocked the port).
I understand completely you not changing the software in this circumstance. Is it possible that the 10 minute reflector check interval is exposed in a parameter list somewhere that I could modify as a test?