Over the past couple of days I've see some odd behavior with WeatherCloud updating / observation posting. I suspect they changed something which slowed down the reply to an update. I haven't found anything about the protocol by the way, but it certainly looks like HTTP from the Diags log.
At any rate, sometime around 2018-01-23 11:20 (Eastern US), it looks like the reply time from WeatherCloud increased noticeably. Before that it looks like it was running around 30 seconds. Afterward it appears to have jumped to over 1:40 (or 100 secs), what apparently generates this message as the task waiting for the reply times out. (Same timing for WU it appears.) This seemed to occur at every attempt for about 19 hours.
This morning, around 2018-01-24 06:40 (Eastern US), it looks like the reply time dropped just enough to avoid this message ... but with very little margin. The margin looks to be on the order of 10 sec or less, or > 10% if you will. A portion of the Diags log, grepped for WeatherCloud follows below.
In general, it seems like the observations were getting through, so the delay could just be on the confirmation handshake.
I did report this to WeatherCloud but haven't heard anything back ... though maybe the change in timing this morning is a reply of sorts.
Code: Select all
20180123-152519.txt:2018-01-24 06:21:40.981 WeatherCloud update: The Task was canceled
20180123-152519.txt:2018-01-24 06:31:41.021 WeatherCloud update: The Task was canceled
20180123-152519.txt:2018-01-24 06:41:34.019 WeatherCloud Response: OK: 200
20180123-152519.txt:2018-01-24 06:51:40.075 WeatherCloud update: The Task was canceled
20180123-152519.txt:2018-01-24 07:00:24.112 WeatherCloud Response: OK: 429
20180123-152519.txt:2018-01-24 07:11:38.906 WeatherCloud Response: OK: 200
20180123-152519.txt:2018-01-24 07:21:36.667 WeatherCloud Response: OK: 200