502 errors are usually temporary, or potentially processing timeouts depending on the endpoint you’re using. It’s not unusual to see the occasional 502 or 504 error after a large number of requests (in the range of tens of thousands of requests over a sustained period), or if requesting or updating too many records in a single request using any of the batch APIs.
Do you have an example of an endpoint that you saw this with?
@dadams IQAir was the one where we were seeing the issue.
Scribe’s reply is here…
First from my own searching on the 502 Bad Gateway error the most common cause is a down server. The second possibility is a network issue that can look like the server is down but instead is inaccessible given the network issue. In this case, given it went on for a couple days, and each time processed many records before failing, I am suspecting a network issue. Also, if Hubspot’s servers had been down for two days we all would have heard about it via any number of channels. Second I agree it would have been better if the 502 Bad Gateway error had been presented in the Execution History. I have circulated this specific issue internally, it resulted in a fair bit of email traffic and is being considered for future releases of Granite. Overall we are looking at the Logging in Granite with the intent of making Logs and Execution History more informative in future releases. Thanks for your comments.