Well Jill’s spots were well-formed but there was a 25min window when she was spotting when the SW3 API was failing. The timeout for comms between the spottter and the API is 10secs. It looks the API was taking in excess of 10secs at this point. It normally takes about 1sec to respond.
I can fix the symptom by extending the timeout to something longer and maybe that would be enough. But that is only fixing the symptom: slow response under some condition. As it normal comes back almost at once this does feel like papering over the cracks. It would be better to understand why the API was slow. It’s not slow now. The API is a cloud instance so maybe it was raining somewhere in Azureland
I haven’t seen lethargic API response for months and months, I’ll keep an eye open.